Inflation as an emergent phenomenon

Inflation as an Emergent Phenomenon

Why do some economies spiral into rapid price increases while others remain stable despite similar external shocks? For decades, economists have treated inflation as a predictable outcome of aggregate forces. These include central bank interest rates or total demand. Most models assume a "representative agent" (a theoretical average person or firm) reflects the whole economy. However, this top-down view struggles to explain why inflation can be sudden, persistent, or localized to specific industries.

A new study from the University of Catania and Marche Polytechnic University proposes that inflation is an emergent property. Much like how traffic jams emerge from the individual braking decisions of drivers, the authors argue that inflation arises from decentralized interactions. It stems from the messy behavior of unique firms, banks, and supply chains.

Beyond the Representative Agent

Traditional macroeconomic models often rely on the Phillips curve or similar aggregate frameworks. These models assume that if you know total demand and labor costs, you can predict inflation. This approach treats the economy as a monolithic block. It ignores the granular reality of how products are actually made and sold.

The authors argue this falls short because it overlooks the "composition problem." In a real economy, prices are not adjusted uniformly. Some firms might be slow to change prices (price stickiness). Others react instantly to cost changes. Furthermore, the way a shock to the price of steel travels to the price of a car depends on specific supply chain connections. As shown in the baseline validation, a stable economy is not a static state.

It is a delicate balance of continuous, uncoordinated transactions. When this balance shifts, inflation becomes a statistical summary of millions of individual decisions.

The Architecture of a Pricing Cascade

To capture this complexity, the researchers developed an agent-based model (ABM). This is a bottom-up simulation where individual agents interact. The model uses a Stock-Flow Consistent (SFC) framework. This ensures every dollar spent by one agent is recorded as income for another. This prevents "financial black holes" where money disappears from the accounting. The model operates through four interconnected layers:

1. Production Networks: Firms are linked in a directed graph. A firm producing consumer goods must purchase intermediate inputs (raw materials or parts) from upstream suppliers. These links create a "pass-through" mechanism. If an upstream supplier raises prices, the downstream firm faces higher unit costs.
2. Endogenous Money and Credit: Here, banks create deposits by lending to firms. This means credit availability and interest rates are tied to actual production needs.
3. Decentralized Pricing: Each firm sets its price using a "cost-plus" rule. They take their unit costs and add an idiosyncratic mark-up (a profit margin). They then adjust based on their own expectations of future prices.
4. Adaptive Learning: Instead of having perfect foresight, firms use "bounded rationality" (making decisions based on limited information). They observe market-level price signals and use adaptive heuristics (rules of thumb) to guess future trends.

These layers interact to create feedback loops. For example, a rise in interest rates does not just dampen demand. It increases the cost of working capital (money used for day-to-day operations). This can push prices higher through the cost-plus pricing rule.

Identifying the Drivers of Price Pressure

The researchers conducted various simulations to isolate which mechanisms drive inflation. The results reveal that not all "shocks" are created equal.

The authors report that "mark-up pressure" is the strongest inflationary driver. This occurs when successful firms increase their profit margins. However, this comes at a significant cost. It is accompanied by a sharp contraction in real economic output . In contrast, financial-cost shocks and changes to the policy rate generate more moderate, "cost-push" inflation [, Figure 4].

These shocks raise inflation by increasing the cost of borrowing.

The study also highlights the importance of the production network's topology (its structural shape). The authors find that increasing "downstream exposure" strengthens the pass-through of costs .

This means more consumer-goods producers rely on intermediate inputs. Finally, the paper demonstrates that expectations act as an amplifier. The authors find that expectations do not create inflation from nothing. Instead, they significantly prolong and intensify price movements once a real cost pressure is already present .

Limits of the Network View

While the model provides a sophisticated lens, it has limitations. The authors utilize Leontief-type input requirements. This means they assume fixed proportions of inputs are needed for production. In a real economy, firms can often substitute one input for another. This flexibility is not currently captured in the model.

Furthermore, the model is a simulation of "regimes" rather than a predictive tool for specific dates. It shows how inflation can happen, but it does not predict exactly when a specific country will experience it. For a practitioner, the model is better suited for stress-testing policy ideas. It helps answer questions like "what happens to inflation if credit markets tighten?"

The Verdict: A Diagnostic Shift

The findings suggest that a "one-size-fits-all" approach to monetary policy may be flawed. If inflation is an emergent phenomenon, a uniform increase in interest rates might be a blunt instrument. It could miss the mark or even worsen the problem.

The authors' results indicate that if inflation is driven by upstream bottlenecks, a rate hike might be counterproductive. Such a move could inadvertently increase production costs for firms already struggling to supply the market. Therefore, the paper argues that central banks should avoid reacting solely to aggregate CPI (Consumer Price Index) numbers. Instead, they should adopt a "diagnostic" approach. They should monitor sectoral indicators, credit spreads, and supply chain bottlenecks. This helps identify the actual source of pressure before choosing a response.

Figures from the paper